Excellent article along with great sound sample files to corroborate.
Excellent article along with great sound sample files to corroborate.
Does Monster™ make interconnects for ear-buds? Something with oxygen-free copper … or gold-plated crystal lettuce … about the diameter of a garden-hose … robust in construction … unidirectional … lubricated with high viscosity snake-oil … but as light as a feather? Because, until Monster™ makes iPod ear-bud cables of that quality … I see no reason for 24/192 audio files … IYKWIMAITYD.
I agree for the lack of need. I will still record at 24/96 which happens to be one of Dolby’s true HD spec the other being 24/192.
Thanks for sharing the link, Larry. I love the science within the article.
Yeah the science is what attracted me since I was and always will be a researcher type (first job was at IBM’s T.J. Watson Research Center). (Interesting story: I once had to fix Dr. Mandlebrot’s PC since he worked 2 aisles over. )
The interesting thing to note is that, if you are using a fully analog setup, then the higher sample rates are necessary. But since most of us are using digital for mixing plus signal processing (inserts / sends) then 44.1 is more than sufficient due to the Nyquist theory. Note how he also addresses the 16- vs. 24-bit depth and says basically what we all know: 16 is fine for playback (44.1/16 = CD after all) but 24 allows the mixing engineer to “set it and forget it” since the noise floor is low enough to allow for midrange levels thus eliminating concerns about clipping during tracking.
thanks for the link…very informative
One thing to note (I can’t remember if this was mentioned in the article) was that this discussion is about the distribution file, not the recording format. I found this interesting article from Variety of Sound regarding ITB mixing and benefits from higher sample frequencies.
Great article! As I’ve always said: 44.1/16 is more than enough for delivery medium. Don’t get me wrong. I record in 88.2/24, because that’s lowest standard suitable for processing IMO.
Excuse me, but where do you need sampling rate, if everything’s analog?
OMG! This guy is genious! In less than 30 minutes he busted all the digital audio myths from sample rate to bit depth and even dithering! And he even used a cheap and obsolete converter to do the job!
Thanks for catching that. I wrote that comment from memory of what the article said. Here’s the exact part of the article that I was mentally trying to comment on:
All signals with content entirely below the Nyquist frequency (half the sampling rate) are captured perfectly and completely by sampling; an infinite sampling rate is not required. Sampling doesn’t affect frequency response or phase. The analog signal can be reconstructed losslessly, smoothly, and with the exact timing of the original analog signal.
So the math is ideal, but what of real world complications? The most notorious is the band-limiting requirement. Signals with content over the Nyquist frequency must be lowpassed before sampling to avoid aliasing distortion; this analog lowpass is the infamous antialiasing filter. Antialiasing can’t be ideal in practice, but modern techniques bring it very close. …and with that we come to oversampling.
Sampling rates over 48kHz are irrelevant to high fidelity audio data, but they are internally essential to several modern digital audio techniques. Oversampling is the most relevant example .
Oversampling is simple and clever. You may recall from my A Digital Media Primer for Geeks that high sampling rates provide a great deal more space between the highest frequency audio we care about (20kHz) and the Nyquist frequency (half the sampling rate). This allows for simpler, smoother, more reliable analog anti-aliasing filters, and thus higher fidelity. This extra space between is 20kHz and the Nyquist frequency is essentially just spectral padding for the analog filter.
What I was inferring was that the analog filter referred to is outboard gear that is used during mixing or mixdown.
That was a really good video, thanks!
Thanks Larry. This article was posted over at Gearslutz a while ago; generated a lot on interesting discussion based on misinformation! Well it that isn’t what forums are for, I don’t know what we’re doing here!
I’m siding with Nyquist at this point. When CD’s first came out (before I was born), there was a lot of argument about how harsh they sounded, presumably due to everything being 1’s and 0’s, and therefore, having “edges” to it. In reality, the first CD’s exposed high and low frequencies previously not audible on LP’s, and that was the harshness audiophiles were hearing. Or so my grandfather tells me. I don’t think there’s anything more available once you pass 44Khz, or at least I am unable to hear it.
Find a few of those ear/equipment tests with tones you can listen to between 5kHz-20kHz and you’ll notice the high end drops off before you want it to, unless you’re a toddler or in best cases a teenager. Also you don’t perceive it like you “hear” the highest overtones around 15kHz but you “feel” them? At least that’s what it seems to me. At the same time those sounds are very faint
Now take that up an octave which is what doubling the frequency means and tell me it really matters …
Maybe indirectly the intermodulation of overtones might have some subtle effect maybe even important???
Naw!!! Screw it! I’ll continue with my faulty dull crappy 44.1/24 recordings … and I’m proud of it!
Well, there are too many moving parts to say exactly what this is, but when we were doing the Sandy project I had to render some MP3s for the first radio show. What I did initially was render to WAV then use the built-in MP3 encoder in WinAmp to quickly convert the file.
But when I listened to it, it sounded like crap. I was on the phone with Tom about this and as I was discussing it with him, I rendered another version using the encoder built into Cubase, loaded both MP3s into Cubase and used Blue Cat Audio’s Frequency Analyst plug-in to see what was happening.
The WinAmp file had an extremely noticeable cutoff at 20kHz. It’s almost like they ignored any frequency above that in a FFT and just didn’t render them in the MP3. The Cubase MP3 had a gentle slope off as if it were preserving the original frequencies.
Now I don’t know if it were the missing frequencies that were causing the very audible issues or if those frequencies were artificially introduced harmonics that resulted as a result of better processing in the audible bands. But whatever it was the difference was huge.
I know I’ve promised a few A/B comparisons for other things, but I’ll try to have up an example of this later this morning.
Ok maybe I was on drugs or it was just the compositions of the songs but the screenshots below don’t corroborate my “it was a huge difference” statement. Having said that, here is what you are looking at below.
I used It’s a Sunny Day as a sample.
I took three sections: measures 1-9 (soft ramp up); measures 41-43 (big horns, snare, and cymbals); and measures 105-109 (big horns but not much else)
I used Blue Cat’s Frequency Analyst with no peak reset so the screenshots below show the maximum amplitude at each frequency band. Other than that, the default settings were used
I took screenshots of the sections within the native project (suffix Native, below). Then I exported a 44.1 / 16 WAV (suffix WAV) and a 192kbps MP3 (suffix Cubase). Finally, I converted the WAV to an MP3 using the built-in WinAmp converter (suffix WinAmp).
Note that the differences between the native Cubase screenshot and WAV file are minimal. This shouldn’t be surprising, but since Cubase stores things internally differently (my project settings are 44.1 / 24-bit, and we all know that really it’s stored as a 32-bit float) I wasn’t expecting things to be 100% identical.
Larry, try using sonogram/spectrogram analysis. It is much more revealing than a curve of a spectrum analyzer.
Got any good plugins that are either free/shareware or cost less than, say, $30? I’ll pay assuming you can tell me why I need to use one in my workflow since I’m ignorant and don’t already know why.
I think MP3 conversion is a completely different subject of psycho-acoustic tricks, and shouldn’t be mixed up with the 96Khz discussion.
Nice, thanks, bookmarking.